The Journal of Space Syntax

ISSN: 2044-7507 Year: 2012 volume: 3 issue: 2 Online Publicaton Date: 28 December 2012
htp://www.journalofspacesyntax.org/
J
S
S
Scaling relative asymmetry in space syntax analysis
Mrio Krger Andrea Pera Vieira
Professor of Architecture Faculty of Architecture
University of Coimbra, Portugal University of Porto, Portugal
Pages: 194-203
194
J
O
S
S
Scaling relative asymmetry in space syntax analysis
Mrio Krger Andrea Pera Vieira
Professor of Architecture Faculty of Architecture
University of Coimbra, Portugal University of Porto, Portugal
1. Introducton
At the social level, space affects human behaviour
and has the potential to induce our actions and infu-
ence their usages. The space syntax theory sup-
ports the idea that space and its confguration have
a great infuence on the socialisation processes that
occur in those occupied and used spaces.
This theory was frst developed at University
College London in the Unit of Architectural Studies
(Hillier and Hanson, 1984) and has a particular way
of representing space in order to systemise infor-
mation for the comprehension of different spatial
characteristics.
The space syntax studies developed at Universi-
ty College London have led to the natural movement
theory, concluding that through the combination of
different information about the spatial patterns and
observation studies, pedestrian movement tends to
be associated with the morphology of the space. In
other words, space syntax states that some places
are better integrated than others, usually indicated
by a higher fow of people. This type of relationship
does not depend only on the individual spaces, but
on the confguration of those spaces as a whole
(Hillier et al., 1993).
This paper reports on a study of space syntax measures and focuses on the standard deviation of the
depth from an axial map. The frst section of the paper is a partial review of the original study On node
and axial maps: Distance measures and related topics (Krger, 1989). The following sections present
new developments whereby a more robust statistical approach to work with integration is used, which not
only considers the mean values given by Relative Asymmetry (RA), but also the corresponding standard
deviation. In other words, the proposition is to work not only with a measure of centrality (1/RA), but also
with a dispersion measure in order to obtain a more complete picture of the distribution of depth in an axial
map. The result of this study on space syntax measures takes into account the standard deviation of the
depth from an axial map, proposing a new measure of Scaled Relative Asymmetry of axial line i (SRA
i
),
which suggests powerful correlations with natural movement.
Keywords:
Space syntax
measures, axial
map analysis, depth
standard deviation.
The theory of space syntax aims to analyse
space and its configuration, focusing on their
implications for social relations and pedestrian
movement. This method allows the study of differ-
ent systems of spatial relations, which characterise
different spaces (Hillier and Hanson, 1984; 1987).
Spatial systems are graphically represented by
their axial map the bi-dimensional representa-
tion of the main lines that connect the entire spatial
system, in which every line stands for a possibility
of fow between two spaces without physical or
visual barriers.
Interpretation of space syntax measures in
these maps is sensitive to the scale of the maps,
since their values are dependent on the size of the
space under study. This issue is particularly relevant
when we compare measures across different urban
or buildings spaces, and it is therefore necessary
to place variables on a common scale obtained
by standardisation methods. This standardised
measure, introduced by Hillier and Hanson (1984)
for expressing integration, is called Real Relative
Asymmetry (RRA).
195
J
O
S
S
Scaling relatve
asymmetry
Krger, M. & Vieira, A.
Krger (1989) also indicates a standardisation
procedure for RA Real Relative Asymmetry (RRA)
that is presented in the next section of this paper.
Normalisation is obtained by comparing a centrality
measure of a node of a graph with n nodes, with the
centrality measure we would get if that node were
the root of a standardised graph in a diamond shape
with the same number of nodes.
These procedures have been shown to be
robust in practice, but nevertheless have been a
matter for considerable discussion. Refecting on
the problem of desirable integration measures that
are independent of the size of the axial map of
urban or building space, Teklenberg, Timmermans
and Wagenberg (1993) propose a new measure
and compare it with the existing measures of RRA,
suggesting a logarithmic transformation of the total
depth of a system. However, this method cannot
produce values for all axial maps if there is a space
where total depth is less than or equal to the total
number of spaces in the system. These authors sug-
gest an integration score primarily for urban plans
or very large buildings and, in the other cases, the
distribution of integration should be calculated using
Hillier and Hansons (1984) method.
As a matter of fact, the Teklenberg, Timmermans
and Wagenberg (1993) approach relies on the
standardisation of mean integration but does not
take into account the standard deviation of depths
values. Also Conroy-Dalton and Daltons (2007)
work assumes a decay function for the distribution
of depth values, making an hypothesis on the form
that distribution, which is not necessary if we have
mean and standard deviation of d-values to com-
pare two or more distributions.
Indeed, mean and standard deviation values are
essential for understanding the distribution of space
syntax values in axial maps because, regardless
of the mean, it makes a great deal of difference
whether the distribution is spread out over a broad
range or clustered closely around the mean.
The work presented here is based on the study
On node and axial maps: Distance measures and
related topics (Krger, 1989), and also on new
developments which consider a more robust statisti-
cal approach to work with integration that not only
takes into account the mean values given by RA, but
also the corresponding standard deviation. Con-
sequently the paper has two parts, the frst being
a review of the original study (ibid.) which presents
some basic space syntax measures and the deri-
vation of the RA measure, based on mean depths
from an axial line to all others. In the second part of
the paper, a new measure called Scaled Relative
Asymmetry (SRA) is developed which aims to take
into account not just mean depths, but also a meas-
ure of their variation. The proposition is therefore to
not only take into account a measure of centrality
(1/RA), but also a dispersion measure in order to
obtain a more complete picture of the distribution
of depths in an axial map. Subsequently it is sug-
gested that SRA performs better then RA, since it
takes into account the form of depths distribution
and not just its mean.
2. General Propertes of Axial Maps
Axial maps usually represent different properties of
urban form and consist of the fewest longest straight
lines that cover all urban public spaces, i.e. lines that
pass through all urban public spaces confgured as
unifed places. These axial lines have properties
of visibility, referring to how far one can see; and
permeability, relating to how far one can go.
However, a more precise defnition is needed if
we want to achieve an accurate description of these
maps in order to explore their properties.
An axial map (AM) consists of a fnite non empty
set L = L(A) of k lines together with a prescribed set
X of m unordered pairs of lines of L.
Each pair x = {u, v} of lines in X is called a con-
nection (or point) and x is said to join u and v. We
196
J
O
S
S
The Journal of
Space Syntax
Volume 3 Issue 2
It should be noted that the application AM (m, k)
GM (k, m) is non isomorphic; i.e. while an axial
map corresponds to just one graph, to the same
graph there correspond many axial maps. In short,
an axial map AM (m, k) corresponds to one graph
G (k, m), but the converse is not true.
For an axial map, the maximum number of con-
nections for a given set of k lines is given by (C
k
2
)
(1)
which is identical to the maximum number of lines
that a graph G with k points can have (Harary, 1971,
p.16). In graph theory terminology, G is called a
complete graph since every pair of its k points is
adjacent. In a similar way we can say that an (m
max
,
k) axial map is a complete axial map.
A graph is said to be connected if every pair of
points can be joined by a path; i.e. by an alternating
sequence of points and lines, in which all points and
lines are distinct and where each line is incidental
write x = uv and say that u and v are adjacent axial
lines; point x and line u are incidental to each other,
as are x and v.
An axial map with k lines and m connections is
called a (m, k) map, the (0, 1) map being a trivial
case represented just by an axial line.
For the (8, 6) axial map represented in Figure 1,
the set of lines is defned as being given by L = {1,
2, 3, 4, 5, 6} and the set of connections as being
given by X
1

= {1, 2}, X
2

= {2, 3}, X
3

= {3, 4}, X
4

= {3,
6}, X
5

= {3, 5}, X
6

= {4, 6}, X
7

= {4, 5} and X
8

= {1, 6}.
A graph G of a (m, k) axial map consists of a
fnite non-empty set V = V(G) of k vertices together
with a prescribed set X of m unordered pairs of
distinct vertices of V. Each vertex in G represents
a line of the (m, k) axial map and each pair y = {r,
s} of vertices in G represents a connection of the
axial map. Each pair y = {r, s} of vertices in G is an
arc of G and y is said to join u and v. A graph G
with k vertices and m arcs is called a (k, m) graph.
The (6, 8) graph represented in Figure 1 is de-
scribed by the set of vertices V = {1, 2, 3, 4, 5, 6}
and by the sets of arcs Y
1

= {1, 2}, Y
2

= {2, 3}, Y
3

=
{3, 4}, Y
4

= {3, 6}, Y
5

= {3, 5}, Y
6

= {4, 6}, Y
7

= {4, 5}
and Y
8

= {1, 6}.
AM (m, k) GM (k, m)

m
max
=
k(k 1)
2

Figure 1:
An example of an (8, 6)
axial map and its corre-
spondent (6, 8) graph.
197
J
O
S
S
Scaling relatve
asymmetry
Krger, M. & Vieira, A.
to the two points immediately preceding and follow-
ing it. A path is considered closed if its frst point is
identical to the last one. For a minimally connected
axial map, the corresponding graph G is called a
tree; i.e. a connected graph with minimum number
of lines, without closed paths or cycles. In a tree
with k vertices there must be k-1 lines; thus a lower
limit (m
min
) for the number of connections in the axial
map is given by k-1.
3. Defniton of Distance Measures on Axial Maps
Several distance measures have been proposed
in the literature to analyse the performance of the
graph representation of the axial map.
In general, we can speak of the distance d
ij

between two points i and j in graph G as being
the length of the shortest path joining them, if any;
otherwise d
ij
= . In a connected graph, distance
presents metric properties, i.e. for all points i, j and
k (Harary, 1971, p.14), the following set of axioms
holds:
1. d
ij
0 , with d
ij
=0 if and only if i = j,
2. d
ij
= d
ji
,
3. d
ij
+ d
jk
d
ik
.
In axial maps the distance between line i and
j is, generally measured by the number of depth
steps, i.e. the number of axial lines located on the
shortest path joining them.
Mean depth of line i in an axial map is defned by
(2)
where k represents the number of lines in the axial
map or the number of points in its graph represen-
tation.
Mean depth measures the extent to which a
given line i is segregated from the remaining lines of
each map. In that sense mean depth can be called
a global property of a specifc axial line.

MD
i
=
d
ij
(k 1)
j =1
k

In order to standardise the variation of mean

depth between zero and one, Hillier and Hanson
(1984, p.108) proposed the following measure,
known as the Relative Asymmetry (RA) of a line
or node i
(3)
where the variables have the usual meaning.
To obtain expression (3) we need to know the
maximum and minimum values that an axial line
can have in terms of mean depth.
The minimum value is given when node i in a
graph G is at minimum depth from all other ones,
i.e. when it is at depth 1 from all other nodes. In
that case the minimum mean depth is 1, i.e. MD
min
= 1. In graph theory terminology this corresponds
to the centre of a star.
The maximum value for MD
i
is given when node
i is the end point of a chain, i.e. of a tree with two
points incidental to one line and the remaining (k-2)
points incidental to two lines.

RA
i
=
2(MD
i
1)
(k 2)
Figure 2:
a) Chain having end point with maximum RA.
b) Star having centre with minimum RA.
a) RA
max
b) RA
min
198
J
O
S
S
The Journal of
Space Syntax
Volume 3 Issue 2
The total depth of an end point i in a chain is
given by the following expression ,
i.e. is identical to the summation of natural numbers,
from m = 1, which corresponds to the node j at
depth 1 from i, up to k-1, which corresponds to the
deepest node j from i.
The summation of the series of natural numbers,
from 1 up to k-1, is given by .
Therefore, the mean depth of an end node i in a
chain is given by .
The expression of RA
i
, defned to vary between
0 and 1, is given in its standardised form as
(4)

d
ij
j =1
k

= m
m=1
k1

m
max
=
k(k 1)
2

MD
max
=
k
2

RA
i
=
MD
i
MD
min
MD
max
MD
min
Substituting the values of MD
min
and MD
max

in expression (4), we obtain expression (3) which
gives the value of the RA of point i. Values close
to 1 represent segregated points in relationship to
the whole graph, while values close to 0 represent
points integrated in the system.
However, as it stands, expression (3) does not
allow us to directly compare the values of RA for
points located in maps of different sizes. In fact,
as k increases, the mean depth decreases, cet-
eris paribus, in proportionate terms. This means
that RA measures also decrease in proportionate
terms when the number of axial lines increases;
it is therefore impossible to compare systems of
different sizes.
The usual approach is to compare RA values
for each point with RA values of a root of a diamond
shape. The reason for adopting this procedure rests
on the assumption that, in both cases, the depths
are approximately normally distributed.
Figure 3:
D
46
- Diamond
Shape with 46
points and 9 lev-
els of depth.
Level N Points Depth
- 9 2
0
8
- 8 2
1
7
- 7 2
2
6
- 6 2
3
5
- 5 2
4
4
- 4 2
3
3
- 3 2
2
2
- 2 2
1
1
- 1 2
0
0
199
J
O
S
S
Scaling relatve
asymmetry
Krger, M. & Vieira, A.
A diamond shape, as a graph, is a special form
of justifed graph. A justifed graph is one in which a
point, called the root, is put at the base and then all
points of depth 1 are aligned horizontally above it,
all points at depth 2 from that point above those at
depth 1, and so on until all levels of depth from that
point are accounted for (Hillier and Hanson, 1984,
p.106). In a diamond shape there are k points at
mean depth level, k/2 at one level above and below,
k/4 at two levels above and below, and so on until
there is one point at the shallowest (the root) and
deepest levels (ibid., p.111-112).
For an axial map with k lines, the general pro-
cedure (see ibid., p.112-113) has been to estimate
the D
k
, i.e. the RA of the root of a diamond shape
with k points, and to divide the RA value found for a
specifc line of the axial map by the value obtained
for D
k
. This new value has been called Real Relative
Asymmetry (RRA) in the literature (see ibid., p.112)
and varies above and below 1. Values well below
1, such as those lower than 0.6, indicate strongly
integrated lines in the axial map, whilst values above
1 indicate more segregated lines.
4. Scaling Relatve Asymmetry
In order to compare the performance of different
procedures to standardise the RA of an axial map,
we need to obtain an expression for the RA of the
diamond root as a function of the number of its
points.
In a diamond shape with k nodes, the total depth
from its root TD
k
, in relationship with all other points,
is given by the following expression
(5)
where d represents the maximum depth from
the root and q the depth, also from the root, of the
points located on each level.

TD
k
= q(2
q
q=0
d / 2

) + (d q)(2
q
q=0
d / 21

)

S
1
= q(2
q
q=0
d / 2

)
The frst term on the right hand
side of equation (5) represents the total depth of
the root in relationship to those points located from
depth 0 to depth d/2. The second term
represents the same in
relationship to those points situated at depth (d/2+1),
up to maximum depth d from the root.
As, in general,
(see Graham et al., 1989, p. 33), then the frst term
in the right hand side of expression (5) becomes
.
For the second term, after expansion, it beco-
mes . .
Substitution of these expanded terms S
1
and S
2

in (5) gives the following result
(6)
The frst two terms in the right hand side of
expression (6) partially cancel out, giving the fol-
lowing result
(7)

But, in general, as then,
developing the second term in the right hand side
of expression (7), we obtain
(8)

S
2
= (d q)(2
q
q=0
d / 21

)

q(2
q
q=0
n

) = (n 1)(2
(n+1)
) + 2 [ ]

S
1
= (
d
2
1)(2
(
d
2
+1)
) + 2
[ ]

S
2
= d(2
q
q=0
d / 21

) + q(2
q
q=0
d / 21

)

TD
k
= S
1
+ S
2
= q(2
q
q=0
d / 2

) + q(2
q
q=0
d / 21

) d(2
q
q=0
d / 21

)

TD
k
= (
d
2
)(2
d / 2
) + d(2
q
q=0
d / 21

)

ax
k
k=0
n

=
(a ax
n+1
)
(1 x)

TD
k
= (
3
2
)(d2
d / 2
) d
200
J
O
S
S
The Journal of
Space Syntax
Volume 3 Issue 2
Expression (8) gives the total depth of a root of a
diamond shape as a function of d, i.e. as a function
of the maximum depth from that root.
As in a diamond shape (see Figure 3) d/2=n,
where n in expression 2
n
represents the depth of the
diameter of a diamond, i.e. of the diamonds level
with the greatest number of points, and 2
n
stands
for the number of points at that level, then if we
substitute this result in (8) we obtain, after algebraic
manipulation, for the total depth of a diamond
(9)
Expression (9) gives the total depth of a diamond
root as a function of its diameter depth.
The total number k of points in a diamond shape
can be given as a function of its diameter depth,
i.e. as a function of n by the following expression


where the frst expression on the right hand side
represents the number of points at diameter level
and the second one the number of points at all
other levels.
As, in general, then, after
algebraic manipulation, the last equation for k could
be transformed, by substitution, into
(10)
If we substitute (10) in (9) we obtain an expres-
sion for the root total depth as a function of the
number of diamond points (k) as well as a function
of its diameter level (n), i.e. simply as
(11)
Then the mean depth of a diamond root can
now be given by
(12)

TD
k
= 2n(3 2
(n1)
1)

k = 2
n
+ 2 2
i
i =0
n1

2
i
i =0
n

=
(1 2
n+1
)
(1 2)

k = 3 2
n
2
If we substitute expression (12) in (3) we obtain
the RA of a root of a diamond (D
k
) as a function of
the number of points k and the depth of its diameter
n, i.e. by
(13).
However, from expression (10) we can estimate
n as a function of k, which is given by
(14).
If we substitute the value of n, given by ex-
pression (14), in (13) we fnally obtain the RA of a
diamond root simply as a function of the number of
its k points, i.e. as
(15).
The usual procedure in space syntax analysis
is to standardise the RA by the values given by
D
k
,
regardless of the form of depths distribution
in axial maps.
If we want to compare axial maps of different
sizes then we should have the same yardstick - not
just in terms of their mean depths, but also concern-
ing the dispersion of their values. In short, we should
account for the entire distribution of depths on maps
to be compared, not just their mean depths, as hap-
pens in the estimation of D
k
which is only dependent
on the value obtained for MD
k
.
Therefore, we now need to obtain the standard
deviation of depth values from a root of a diamond
shape in order to convert RA values to a common
scale and be able to compare axial maps of dif-
ferent sizes.
The following identity provides a basis for esti-
mating the standard deviation of variable x on an
interval scale distribution:

TD
k
= k n

MD
k
= (k n) /(k 1)
D
k
=
2 k(n 1) +1
[ ]
(k 1)(k 2)

n = lg
2
k + 2
3






D
k
=
2 (k(lg
2
(
k+2
3
) 1) +1)
[ ]
(k 1)(k 2)
201
J
O
S
S
Scaling relatve
asymmetry
Krger, M. & Vieira, A.
(16)
where k represents the number of observations,

the mean of the distribution and MD
k
, as usual, the
mean depth of a diamond root.
In order to estimate the standard deviation of
depths on a diamond root ( D
k
), we need to express
in context, the expression ; i.e. we need to
develop it in a similar fashion as we did for the total
depth from its root (TD
k
) given by expression (5).
In other words, we need to estimate the follow-
ing expression,
(17)
where the variables have the usual meaning.
As the diameters depth of a diamond shape
equals half of the maximum depth from the root (n =
d/2), then the right hand side of equation (13) could
be transformed, by substitution, into
(18).
Developing and factoring both terms of expres-
sion (18), we obtain,
(19).
If we substitute expressions (12) and (19) in
(16), we fnd the standard deviation of depths on a
diamond root ( D
k
) given by
(20).

(x
i
x)
2
i =1
k

k1








=
x
i
2
i =1
k

k1
+ (k 2) MD
k
( )
2









x
i
2
i 1
k

x
i
2
i=1
k

= q
2
(2
q
q=0
d / 2

) + (d q)
2
(2
q
q=0
d / 21

)

q
2
(2
q
q=0
n

) + (d q)
2
(2
q
q=0
n1

)

x
i
2
i=1
k

= (3 2
n
n
2
4n
2
8n + 3 2
n+2
12)

D
k
=
3 2
n
n
2
4n
2
8n + 3 2
n+2
12
k 1
(k 2)
kn
k 1






2
As , we can express the D
k

just in function of its k elements, as we did it for D
k
.
This gives a general procedure to standardise
depth measures in order to convert them with a
certain mean and standard deviation to a com-
mon scale that is only dependent on the number
of their nodes.
That transformation can be done in one step by
the following equation that converts values in one
scale directly to comparable values in another scale
by means of a linear transformation (Guilford and
Fruchter, 1978, p.477)
(21)
where ;
, where d
ij
stands for the
shortest path between nodes i and j;
;
D
k
is the standard deviation of depths for the root
of a diamond shape with k elements;
D
i
is the standard deviation of depths for axial
line i;
and STD
i
represents the standardised total depth
of axial line i.
Knowing D
i
and TD
i
from a particular distribu-
tion of depth values in an axial map, we are now able
to obtain the standardised value of its total depth
STD
i
and therefore the Scaled Relative Asymmetry
of axial line i (SRA
i
) given by its scaled mean depth
(SMD
i
) as
(22),
where SMD
i
= STD
i
/(k-1); SMD
max
represents the
standard mean depth of the end point of a chain

n = lg
2
((k +2) / 3)

MD
k
= (k n) /(k 1)

MD
i
= d
ij
j =1
k

/(k 1)

TD
k
= 2n(3.2
n1
1)

SRA
i
=
SMD
i
SMD
min
SMD
max
SMD
min

STD
i
=
D
i
D
k






TD
k

D
i
D
k






MD
k
MD
i






202
J
O
S
S
The Journal of
Space Syntax
Volume 3 Issue 2
with k nodes, and SMD
min
the standard mean depth
of the centre of a star with k nodes.
The expression (22) is equivalent to RA
i
for
non-standardised depth values that were defned to
vary between 0 and 1 and given by expression (4).
We now need to estimate SMD
min
and SMD
max

in a similar fashion as we did for STD
i
using an ex-
pression, in both cases, analogous to equation (21).
However, as the centre of a star has standard
deviation equal zero, it means that the standardisa-
tion procedure given by the expression
(23),
where, STD
s
stands for the standardised total depth
of the centre of a star with k nodes, D
s
for standard
deviation and MD
s
for the mean depth of that node,
and the other variables have the usual meaning,
then it becomes STD
s
= MD
s
(because D
s
= 0).
Therefore, the standardised mean depth (SMD
s
)
of a stars centre with k nodes transforms into
(24),
On the other hand, to obtain SMD
max
we need
to calculate STD
c
, i.e. the standardised total depth
of the end point of a chain with k elements given by
(25),

SMD
min
= SMD
s
=
STD
s
k 1

STD
s
=
D
s
D
k






TD
k

D
s
D
k






MD
k
MD
s







SMD
s
=
1
(k 1)

SMD
Max
= SMD
c
=
STD
c
k 1

STD
c
=
D
c
D
k






TD
k

D
c
D
k






MD
k
MD
c






where D
k
and D
c
are, respectively, the standard
deviations of depths for the root of a diamond shape
and of the end point of a chain with k elements;
; MD
c

= k/2 stands for the
mean depth and for the total depth from the root
of a diamond shape with k nodes.
The only term that needs now to be estimated
is D
c
. That can be done in a similar fashion as
we did to obtain the standard deviation of variable
x on an interval scale distribution
= (26).
Substituting the values of D
c
in equation (25),
as well as all other variables already deduced, we
obtain the standardised total depth (STD
c
) of the end
point of a chain with k nodes and, consequently, the
SMD
max
given by
.

We now have fnally all the elements to estimate
the Scaled Relative Asymmetry of axial line i (SRA
i
)
given by equation (22).
Scaling by this procedure assumes that the
obtained form of depth distribution for axial line i is
the same as the original one would be on a scale
of equal units.
As the diamond shape has a distribution of
depths from its root with mean depth MD
k
(equation
12) and standard deviation given by D
k
(equation
20), this enables the scaling of RA by the results
given in equation (22) which, in turn takes into ac-
count not just the values of mean depths, but also
the variability of their distributions in axial maps.

MD
k
= (k n) /(k 1)

D
c
=
(i
k
2
)
2
i =1
k

k1








) 1 ( 12
2
3

k
k k

SMD
max
=
STD
c
(k 1)
203
J
O
S
S
Scaling relatve
asymmetry
Krger, M. & Vieira, A.
5. Conclusions
The frst section of the paper is a partial review of
Krgers original paper (1989), presenting some
space syntax measures which are used in this study.
The results obtained in this paper are the out-
comes of a study that took the standard deviation of
the depth from an axial map into account. Introduc-
ing a new measure for axial analyses, the Scaled
Relative Asymmetry of axial line i (SRA
i
), suggests
more powerful correlations with natural movement.
We propose that testing the correlation of SRA
i
with natural urban movement will produce interest-
ing results. These tests will be undertaken in a future
study of the axial comparative analyses of urban
maps representing Portuguese settlements, as well
as Architecture Faculty buildings in Portugal. In the
latter case, since two-dimensional plane axial maps
do not apply to multi-storey buildings, the diamond
shape should be mapped onto a sphere in order to
take into account the genus of a three-dimensional
surface, while novel derivations should be made
for the mean depth and the standard deviation of
these axial maps.
In short, although the derivation of the expres-
sion for RRA has been available since Krger
(1989), it is possible to work with a more robust
statistical approach which not only takes into ac-
count its the mean depth values, but also the cor-
responding standard deviation for each axial line
given by SRA
i
. In other words, the proposition is to
work not only with a measure of centrality, but also
with a dispersion measure in order to obtain a more
complete picture of the distribution of depth values
in an axial map.
References
Conroy-Dalton, R. and Dalton, N. (2007), Applying depth
decay functions to space syntax network graphs. In:
Proceedings of the Sixth International Space Syntax
Symposium, Istanbul: I.T.U. Faculty of Architecture,
p.089.1-089.14.
Hanson, J. and Hillier, B. (1987) The architecture of
community: Some new proposals on the social con-
sequences of architectural and planning decisions.
In: Architecture et Comportement/Architecture and
Behaviour, Vol. 3 (3), p. 251 273.
Harary, F. (1969), Graph Theory, Addison-Wesley, Read-
ing, Massachusetts.
Hillier, B. and Hanson, J. (1984), The Social Logic of Space,
London: Cambridge University Press.
Hillier, B., Penn, A., Hanson, J., Grajewski, T. and Xu, J.
(1993), Natural movement: Or, confguration and
attraction in urban pedestrian movement. In: Environ-
ment and Planning B: Planning and Design, Vol. 20
(1), p.29-66.
Krger, M. (1989), On node and axial maps: distance
measures and related topics. In: European Confer-
ence on the Representation and Management of Urban
Change, Cambridge: University of Cambridge.
Graham, R. L., Knuth, D. E. and Patashnik, O. (1989),
Concrete Mathematics: Foundation for Computer
Science, Reading, Massachusetts: Addison-Wesley
Publishing Company.
Guilford, J. P. and Fruchter, B. (1978) Fundamental Statis-
tics in Psychology and Education, New York: McGraw-
Hill Book Company.
Teklenberg, J. A. F., Timmermans, H. J. P. and Wagenberg,
A. F. V. (1993), Space syntax: Standardised integra-
tion measures and some simulations. In: Environment
and Planning B: Planning and Design, Vol. 20 (3),
p.347-357.
About the authors:
Mrio Krger (kruger@
ci.uc.pt), is full Professor at
the University of Coimbra,
was a member of the Centre
for Land Use and Built Form
Studies at the University of
Cambridge, which in 1978
awarded him a PhD in the
area of architectural and
urban morphology. He su-
per vi sed more than si x
dozens masters theses
and/or doctoral degrees at
the Universities of Braslia,
Uni versi t y College Lon-
don, Cambridge, Technical
University of Lisbon and
Coimbra. With three books
in the feld of architectural
theory and seven dozens
of papers publ i shed i n
national and international
journals, he produced, as
co-author, a critical edition
in Portuguese of the treatise
De re aedifcatoria by Leon
Battista Alberti, published
in 2011.
Andrea Pera Vieira (apviei-
ra@arq.up.pt) is Assistant
Lecturer in CAAD at Faculty
of Architecture of University
of Porto (FAUP) and is grad-
uated in mathematics and
in architecture at the same
university. She has a Master
in urban planning by the
University of Aveiro (2007)
and is a research student
at FAUP on how learning
spaces can improve learn-
ing activities. She is also a
junior researcher in CEAU
(www.ceau.arq.up.pt) and
participates in research
proj ects on E- Learni ng
Caf and Spatial Represen-
tation and Communication
Center - FAUP.